Skip to content

Posterior probability vs PIP (Posterior Inclusion Probability)

Terms
Posterior Probability Posterior Inclusion Probability (PIP) Fine-Mapping Credible Set
Definition

Posterior probability is the general Bayesian notion: updated probability of a hypothesis or model given data and a prior. PIP is a *specific* posterior quantity in fine-mapping: the probability that a given variant belongs to the set of causal (or driving) variants at a locus under a sparse regression or Bayesian model (e.g. SuSiE).

Topics
Bayesian statistics GWAS

How they differ

Posterior probability (general) PIP
What it quantifies Any event or hypothesis (model choice, parameter, causal variant). Inclusion of a variant in the causal set at a locus under a fine-mapping model.
Comparable to Other posteriors on the same hypothesis space. Other variants’ PIPs at the same locus; used to rank SNPs and build credible sets.
Not the same as A single P value from marginal GWAS. Marginal p values either—PIP conditions on multi-SNP structure and LD.

Rule of thumb: “Posterior probability” is a language umbrella; PIP is the fine-mapping specialization for variant-level inclusion. A paper’s “posterior probability of causality” at a SNP might be reported as PIP when using standard fine-mapping tools.

References

  • Li Z, Zhou X. (2025). Towards improved fine-mapping of candidate causal variants. Nat Rev Genet.